Page 1 of 1

How to Check for Autocorrelation/Serial Correlation?

Posted: Sun Mar 29, 2015 10:55 am
by DogeProphet
I am running a multiple regression trying to work out how HIV affects the GDP of several countries across 24 years. I have 912 observations. My problem is that I don't know how to find out if my data has serial correlation, I suspect it does because my Durbin Watson stat is near 0. I look to confirm this visually but I can't find the correlelogram Q stat option in the software. I'm usig Eviews 8 and I don't know how to find this option. Subsequently, I also need to know how to fix this serial correlation. I was told that I had to run newey-west and re-estimate my model and i don't know how to do this as well. Please Help.

Re: How to Check for Autocorrelation/Serial Correlation?

Posted: Sun Mar 29, 2015 11:36 am
by EViews Gareth
EViews does not report Q-stats in a panel setting.

Re: How to Check for Autocorrelation/Serial Correlation?

Posted: Sun Mar 29, 2015 11:43 am
by DogeProphet
EViews does not report Q-stats in a panel setting.
I seem to remember that Eviews 7 did so. Am i incorrect?
Also how do i fix autocorrelation in my model?
Thanks

Re: How to Check for Autocorrelation/Serial Correlation?

Posted: Wed Aug 19, 2015 11:46 pm
by lauritam
Hi there,

I think you can check for autocorrelation by clicking on View>Residuals Diagnostic>Correlogram-Q-Statistics!

I am still figuring out how to fix it in Eviews!

Good luck